Thanks for your help, guys. Looks like I really have to go for some training sessions, like Martin proposed, since most people tell me, there´s no "right setting".

I´m just wondering, how leaving all volumes at 0 db can work right:

If I understand Dietz, the VSL team kind of optimized all instruments to reach an optimum gain.
So if an ensemble of Horns reaches e.g. a peak of - 3 db and a single Woodwind reaches the same peak, I don´t get, why I should stick to the dynamics of the instruments in order to reach a balanced sound, instead of changing the track volume.

Wouldn´t this mean, that I have a flute playing piano, but with the volume of a forte flute, if I keep the volume at 0 db?

Felix, when someone talks too loud at you, you hear that, right? When someone whispers, you don't hear it much, right? I never change the setting of my output volume so i can have the SAME reference each time, if a flute is too loud it should annoy you, see if you like the flute at that volume, or do you prefer it softer, louder...

Hi Guy,

in this point I don´t agree:

If someone whispers, but I want to hear him louder, I have different possibilites:

a) Ask him to talk louder
b) Get him closer to my ear or giving him a megaphone.

Transferred to the orchestra I still believe, that keeping the volume of the flutes at 0db compares to giving a person a megaphone. This means I hear a sound, that could never been produced in a natural way, cause a single piano flute is not able to reach a certain volume (like a whisper), while alternative a) would mean, asking the flute to play forte instead of piano.

Since I do not have much experience about orchestral music and your stuff in the demo section sounds perfect to me, I´m still stuck at the question, why it works so good in your songs, even though it should not work theoreticly. And why does the VSL Team even record samples of forte flutes, even though producing an "artifical" forte by using the gain, is obviously working fine, cause I don´t think, someone here will doubt, that your stuff sounds very realistic.

Oh sorry, I didn't say that right, what I meant was that if someone is blasting in your ear you won't like that and tell him to speak softer, if someone is whispering you won't hear him well and say, speak up! Maybe you had already understood that, but just in case. Anyway, best of luck! [:D]

just for clarity, i work from a template where the customary orchestral instruments are all set at 0db. the important thing to note is that they have already been balanced to their relative levels by trimming them to the volume the instruments actually sound (e.g. a flute not nearly as loud as a violin section). thus, any volume changes can only be acccomplished within the dynamic range of the instrument and not above where things sound unnatural. there again each instrumental group, strings, brass, winds, percussion, go through a separate submix which adds another layer of volume control. when the final mixes are automated, the levels are ALWAYS moving all over the place just as real instruments do. the point is that, because everything has been trimmed, everything stays balanced properly in relation to each other. the benefits are three-fold. orchestrations follow the normal concepts of orchestration, they transfer accurately to a real orchestra, and the only "mixing" is related to the expression of an individual instrument. this saves enormous time and keeps everything consistent from one cue to the other.

i should also add that much/most of these individual "volume" changes occur via changes in timbre with filter or layer changes that recreate the effect of an instrument getting brighter as it's volume increases. volume changes NEVER happen without timbral adjustment.

so referring to my previous post here, i think about the only way to acheive a template like this is to mock up pieces and reference them against actual recordings. one might be surprised at some of the levels, but the end result is that there won't be an expectation, for example (sorry felix) that adding two oboes to a section of violins would add any discernable volume to the section (they can help define and/or color the musical line but i would add either one or at least three as it generally takes more than two instruments in almost any section to blend well). it is very time consuming to get this set up, but the end result, like a real orchestra is that it constantly reinforces good orchestration habits, saves time and keeps the quality of the final product at its highest possible level.

I agree, and ONLY use volume adjustment at last resources.

I´m affraid, that I´m a helpless case concerning this, cause now my clarity has totally vanished:

@Martin Bayless said:

the important thing to note is that they have already been balanced to their relative levels by trimming them to the volume the instruments actually sound (e.g. a flute not nearly as loud as a violin section).

Did the VSL-Team trimm the levels?
Maybe I expressed my very first question of the thread wrong by asking, how instruments are "recorded". What I wanted to know is, are they trimmed to there volume in relation to all other instruments.

As I understand Martin, they are?! Just to make sure, I´m not getting this wrong again:
If I tell the flute to play a note and the violas to play one note in a real orchestra, the relation of the volume of the two should approxemately be the same as loading a flute and the violas to a 0db EXS-instrument? (Appart from reverb settings, etc.)

On my Computer I get a level of -8 db, if I play a C4 (vel 127) with the HO-4L schmett_3s. Playing a FL1_LV_nA_sus at C4 (vel 127) I get a level of - 7 db. Is this meant to be? How can one flute be as loud as 4 Horns?


One other question to Martin:

By saying "when the final mix is automated", U mean just automating the dynamics and filters, not the volumes of your 4 sub-channels, right?
And the "levels", that are always moving, are not the levels of your volume faders?

Actually, this is, where I want to get with my setup, not needing to touch any volume and do everything by using the layers. But somehow it doesn´t work, like in my example with the flutes and horns.

Hi felix – vsl did not trim the levels. We need to do that on our end. So the theory becomes one of setting the relative loudness of instruments (so that it becomes impossible for example to make a bassoon louder than a trumpet) such that they are generally balanced when all the faders are set to their maximum. Again maybe the best way to get to this point might be to mock up several classical excerpts and reference them against actual recordings. If you do that, it will no longer make any sense to try and do this with decibels.

As far as automated movement of the faders, that happens constantly throughout a piece primarily on the individual instrument track but also to a lesser degree on the group fader as well, this in addition to filter or layer changes. I think in your case felix, that one of your strengths is your sensitivity to volume adjustments within your melodic lines. To that I would only suggest adding (not replacing) the filter/layer idea and working within a template that starts with the instruments in relative balance. Everything beyond that becomes musical sensitivity (which you have) and the finer points involving orchestration and performance practice.

I did not exspect this to happen, but:

I see clear now!

So thanks for all this help!

P.S.: One little question, that´s still unclear:

How do U "trimm"?
The only possibilities, I have in mind, are:

Changing the volume in the trackmixer,
Using a gain-plug-in,
Changing the midi volume.

i think gain plug in might be the direct answer you're looking for as faders and midi volume are better used for expression just as a player would play louder or softer.

there are many ways to get the same job done however depending on how you work. i use the vsl pro edition because i needed to edit patches down to the sample level (not possible with vienna instruments). at that point, i just set each instrument to play at the level it sounds in an orchestra. that must be possible with the vi's just as with gigastudio or kontakt or however you want to bring the samples into your host software without the need to expend cpu resources on a plug in.

I´m starting to believe kotoris crossfading script (if you´re using K2) is a better way of controlling expression. For example, that way the p and f samples are heard at the correct volume (compared to eachother). The p samples should sound soft and the f should sound loud. [:)]

/Johnny